Generative AI impacts our future. Is this our future?
We have a major problem with artificial intelligence and its impact on water and energy consumption levels. Therefore, I thought I would share some key facts regarding ChatGPT’s massive water and energy consumption that all directors and officers need to be aware of:
- ChatGPT consumes more than half a million kWh of electricity every day, enough to meet around 200 million demands.
- ChatGPT’s daily electricity consumption is almost equal to that of 180,000 U.S. homes, each using about 29 kWh.
- A ChatGPT conversation consumes around 50 cl of water.
This is very concerning given the incredible growth of genAI product innovations not only from OpenAI, but also from major technology players or new entrants like: Amazon, Anthropic, Cohere, Microsoft and Nvidia.
The AI sector’s electricity consumption is already expected to increase significantly, reaching between 85-134 TWh per year by 2027.
A recent OECD policy report found that the water footprint of AI varies significantly depending on where it is trained and housed. For example, AI consumes between 1.8 and 12 liters of water for every kWh of energy consumed in Microsoft’s global data centers, with Ireland and Washington State being the most energy-efficient locations, respectively. water and the least water efficient.
We will need much more efficient genAI infrastructures to compress complex AI models, but also to design more energy-efficient and energy-friendly technological innovations. Water is our rarest global resource and purest source of life.
Fresh water represents only 2.5% of the planet’s total volume, more than half of which is ice. Agriculture uses 70% of what is actually usable. By 2050, two-thirds of the world’s population is expected to suffer from water scarcity, affecting every aspect of people’s lives.
Government leaders need to get a better handle on these realities and risks, and directors and CEOs need to put more pressure on tech titans to act more responsibly and ethically when it comes to sustainability. For more information about the OECD, read this newsletter as this strengthens, the general public must become more involved in AI policy issues.
Another article I wrote on this same topic is below:
